Multimodal Search on MAT-Search benchmark leaderboard · SOTA2 Research

Multimodal Search on MAT-Search

86AccuracyGemini-3-Pro

Evaluation Results

Method	Links
Gemini-3-Pro2026.06		86
GPT-52026.06		84.67
SenseNova-MARS2026.06		83.33
Gemini-3-Flash2026.06		82.67
Qwen3-VL-8B-Instruct2026.06		82.67
Gemini-3-Flash2026.06		82.67
TAPO2026.06		82.67
Gemini-2.5-Flash2026.06		82
Gemini-3-Pro2026.06		82
Qwen3-VL-32B-Instruct2026.06		82
Qwen3-VL-8B-Instruct+SAPO2026.06		82
Qwen2.5-VL-32B-Instruct2026.06		81.33
GPT-5.22026.06		80.67
GPT-5.22026.06		80
GPT-4o-mini2026.06		80
GPT-4o2026.06		79.33
GPT-52026.06		79.33
GPT-4o2026.06		76.67
Gemini-2.5-Flash2026.06		75.33
Visual-ARFT2026.06		74
MMSearch-R12026.06		74
Qwen3-VL-32B-Instruct2026.06		72.67
GPT-4o-mini2026.06		72.66
Qwen2.5-VL-32B-Instruct2026.06		71.33
Qwen3-VL-8B-Instruct2026.06		67.33
Qwen2.5-VL-7B-Instruct2026.06		67.33
Qwen2.5-VL-7B-Instruct2026.06		60